ws.c

   1 /*
   2  * Whitespace rules
   3  *
   4  * Copyright (c) 2007 Junio C Hamano
   5  */
   6
   7 #include "cache.h"
   8 #include "attr.h"
   9
  10 static struct whitespace_rule {
  11         const char *rule_name;
  12         unsigned rule_bits;
  13         unsigned loosens_error:1,
  14                 exclude_default:1;
  15 } whitespace_rule_names[] = {
  16         { "trailing-space", WS_TRAILING_SPACE, 0 },
  17         { "space-before-tab", WS_SPACE_BEFORE_TAB, 0 },
  18         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB, 0 },
  19         { "cr-at-eol", WS_CR_AT_EOL, 1 },
  20         { "blank-at-eol", WS_BLANK_AT_EOL, 0 },
  21         { "blank-at-eof", WS_BLANK_AT_EOF, 0 },
  22         { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 },
  23 };
  24
  25 unsigned parse_whitespace_rule(const char *string)
  26 {
  27         unsigned rule = WS_DEFAULT_RULE;
  28
  29         while (string) {
  30                 int i;
  31                 size_t len;
  32                 const char *ep;
  33                 int negated = 0;
  34
  35                 string = string + strspn(string, ", \t\n\r");
  36                 ep = strchrnul(string, ',');
  37                 len = ep - string;
  38
  39                 if (*string == '-') {
  40                         negated = 1;
  41                         string++;
  42                         len--;
  43                 }
  44                 if (!len)
  45                         break;
  46                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  47                         if (strncmp(whitespace_rule_names[i].rule_name,
  48                                     string, len))
  49                                 continue;
  50                         if (negated)
  51                                 rule &= ~whitespace_rule_names[i].rule_bits;
  52                         else
  53                                 rule |= whitespace_rule_names[i].rule_bits;
  54                         break;
  55                 }
  56                 if (strncmp(string, "tabwidth=", 9) == 0) {
  57                         unsigned tabwidth = atoi(string + 9);
  58                         if (0 < tabwidth && tabwidth < 0100) {
  59                                 rule &= ~WS_TAB_WIDTH_MASK;
  60                                 rule |= tabwidth;
  61                         }
  62                         else
  63                                 warning("tabwidth %.*s out of range",
  64                                         (int)(len - 9), string + 9);
  65                 }
  66                 string = ep;
  67         }
  68
  69         if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
  70                 die("cannot enforce both tab-in-indent and indent-with-non-tab");
  71         return rule;
  72 }
  73
  74 unsigned whitespace_rule(const char *pathname)
  75 {
  76         static struct attr_check *attr_whitespace_rule;
  77
  78         if (!attr_whitespace_rule)
  79                 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
  80
  81         if (!git_check_attr(pathname, attr_whitespace_rule)) {
  82                 const char *value;
  83
  84                 value = attr_whitespace_rule->items[0].value;
  85                 if (ATTR_TRUE(value)) {
  86                         /* true (whitespace) */
  87                         unsigned all_rule = ws_tab_width(whitespace_rule_cfg);
  88                         int i;
  89                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  90                                 if (!whitespace_rule_names[i].loosens_error &&
  91                                     !whitespace_rule_names[i].exclude_default)
  92                                         all_rule |= whitespace_rule_names[i].rule_bits;
  93                         return all_rule;
  94                 } else if (ATTR_FALSE(value)) {
  95                         /* false (-whitespace) */
  96                         return ws_tab_width(whitespace_rule_cfg);
  97                 } else if (ATTR_UNSET(value)) {
  98                         /* reset to default (!whitespace) */
  99                         return whitespace_rule_cfg;
 100                 } else {
 101                         /* string */
 102                         return parse_whitespace_rule(value);
 103                 }
 104         } else {
 105                 return whitespace_rule_cfg;
 106         }
 107 }
 108
 109 /* The returned string should be freed by the caller. */
 110 char *whitespace_error_string(unsigned ws)
 111 {
 112         struct strbuf err = STRBUF_INIT;
 113         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 114                 strbuf_addstr(&err, "trailing whitespace");
 115         else {
 116                 if (ws & WS_BLANK_AT_EOL)
 117                         strbuf_addstr(&err, "trailing whitespace");
 118                 if (ws & WS_BLANK_AT_EOF) {
 119                         if (err.len)
 120                                 strbuf_addstr(&err, ", ");
 121                         strbuf_addstr(&err, "new blank line at EOF");
 122                 }
 123         }
 124         if (ws & WS_SPACE_BEFORE_TAB) {
 125                 if (err.len)
 126                         strbuf_addstr(&err, ", ");
 127                 strbuf_addstr(&err, "space before tab in indent");
 128         }
 129         if (ws & WS_INDENT_WITH_NON_TAB) {
 130                 if (err.len)
 131                         strbuf_addstr(&err, ", ");
 132                 strbuf_addstr(&err, "indent with spaces");
 133         }
 134         if (ws & WS_TAB_IN_INDENT) {
 135                 if (err.len)
 136                         strbuf_addstr(&err, ", ");
 137                 strbuf_addstr(&err, "tab in indent");
 138         }
 139         return strbuf_detach(&err, NULL);
 140 }
 141
 142 /* If stream is non-NULL, emits the line after checking. */
 143 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 144                                 FILE *stream, const char *set,
 145                                 const char *reset, const char *ws)
 146 {
 147         unsigned result = 0;
 148         int written = 0;
 149         int trailing_whitespace = -1;
 150         int trailing_newline = 0;
 151         int trailing_carriage_return = 0;
 152         int i;
 153
 154         /* Logic is simpler if we temporarily ignore the trailing newline. */
 155         if (len > 0 && line[len - 1] == '\n') {
 156                 trailing_newline = 1;
 157                 len--;
 158         }
 159         if ((ws_rule & WS_CR_AT_EOL) &&
 160             len > 0 && line[len - 1] == '\r') {
 161                 trailing_carriage_return = 1;
 162                 len--;
 163         }
 164
 165         /* Check for trailing whitespace. */
 166         if (ws_rule & WS_BLANK_AT_EOL) {
 167                 for (i = len - 1; i >= 0; i--) {
 168                         if (isspace(line[i])) {
 169                                 trailing_whitespace = i;
 170                                 result |= WS_BLANK_AT_EOL;
 171                         }
 172                         else
 173                                 break;
 174                 }
 175         }
 176
 177         if (trailing_whitespace == -1)
 178                 trailing_whitespace = len;
 179
 180         /* Check indentation */
 181         for (i = 0; i < trailing_whitespace; i++) {
 182                 if (line[i] == ' ')
 183                         continue;
 184                 if (line[i] != '\t')
 185                         break;
 186                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 187                         result |= WS_SPACE_BEFORE_TAB;
 188                         if (stream) {
 189                                 fputs(ws, stream);
 190                                 fwrite(line + written, i - written, 1, stream);
 191                                 fputs(reset, stream);
 192                                 fwrite(line + i, 1, 1, stream);
 193                         }
 194                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 195                         result |= WS_TAB_IN_INDENT;
 196                         if (stream) {
 197                                 fwrite(line + written, i - written, 1, stream);
 198                                 fputs(ws, stream);
 199                                 fwrite(line + i, 1, 1, stream);
 200                                 fputs(reset, stream);
 201                         }
 202                 } else if (stream) {
 203                         fwrite(line + written, i - written + 1, 1, stream);
 204                 }
 205                 written = i + 1;
 206         }
 207
 208         /* Check for indent using non-tab. */
 209         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
 210                 result |= WS_INDENT_WITH_NON_TAB;
 211                 if (stream) {
 212                         fputs(ws, stream);
 213                         fwrite(line + written, i - written, 1, stream);
 214                         fputs(reset, stream);
 215                 }
 216                 written = i;
 217         }
 218
 219         if (stream) {
 220                 /*
 221                  * Now the rest of the line starts at "written".
 222                  * The non-highlighted part ends at "trailing_whitespace".
 223                  */
 224
 225                 /* Emit non-highlighted (middle) segment. */
 226                 if (trailing_whitespace - written > 0) {
 227                         fputs(set, stream);
 228                         fwrite(line + written,
 229                             trailing_whitespace - written, 1, stream);
 230                         fputs(reset, stream);
 231                 }
 232
 233                 /* Highlight errors in trailing whitespace. */
 234                 if (trailing_whitespace != len) {
 235                         fputs(ws, stream);
 236                         fwrite(line + trailing_whitespace,
 237                             len - trailing_whitespace, 1, stream);
 238                         fputs(reset, stream);
 239                 }
 240                 if (trailing_carriage_return)
 241                         fputc('\r', stream);
 242                 if (trailing_newline)
 243                         fputc('\n', stream);
 244         }
 245         return result;
 246 }
 247
 248 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 249                    FILE *stream, const char *set,
 250                    const char *reset, const char *ws)
 251 {
 252         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 253 }
 254
 255 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 256 {
 257         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 258 }
 259
 260 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 261 {
 262         /*
 263          * We _might_ want to treat CR differently from other
 264          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 265          * for now we just use this stupid definition.
 266          */
 267         while (len-- > 0) {
 268                 if (!isspace(*line))
 269                         return 0;
 270                 line++;
 271         }
 272         return 1;
 273 }
 274
 275 /* Copy the line onto the end of the strbuf while fixing whitespaces */
 276 void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 277 {
 278         /*
 279          * len is number of bytes to be copied from src, starting
 280          * at src.  Typically src[len-1] is '\n', unless this is
 281          * the incomplete last line.
 282          */
 283         int i;
 284         int add_nl_to_tail = 0;
 285         int add_cr_to_tail = 0;
 286         int fixed = 0;
 287         int last_tab_in_indent = -1;
 288         int last_space_in_indent = -1;
 289         int need_fix_leading_space = 0;
 290
 291         /*
 292          * Strip trailing whitespace
 293          */
 294         if (ws_rule & WS_BLANK_AT_EOL) {
 295                 if (0 < len && src[len - 1] == '\n') {
 296                         add_nl_to_tail = 1;
 297                         len--;
 298                         if (0 < len && src[len - 1] == '\r') {
 299                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 300                                 len--;
 301                         }
 302                 }
 303                 if (0 < len && isspace(src[len - 1])) {
 304                         while (0 < len && isspace(src[len-1]))
 305                                 len--;
 306                         fixed = 1;
 307                 }
 308         }
 309
 310         /*
 311          * Check leading whitespaces (indent)
 312          */
 313         for (i = 0; i < len; i++) {
 314                 char ch = src[i];
 315                 if (ch == '\t') {
 316                         last_tab_in_indent = i;
 317                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 318                             0 <= last_space_in_indent)
 319                             need_fix_leading_space = 1;
 320                 } else if (ch == ' ') {
 321                         last_space_in_indent = i;
 322                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 323                             ws_tab_width(ws_rule) <= i - last_tab_in_indent)
 324                                 need_fix_leading_space = 1;
 325                 } else
 326                         break;
 327         }
 328
 329         if (need_fix_leading_space) {
 330                 /* Process indent ourselves */
 331                 int consecutive_spaces = 0;
 332                 int last = last_tab_in_indent + 1;
 333
 334                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 335                         /* have "last" point at one past the indent */
 336                         if (last_tab_in_indent < last_space_in_indent)
 337                                 last = last_space_in_indent + 1;
 338                         else
 339                                 last = last_tab_in_indent + 1;
 340                 }
 341
 342                 /*
 343                  * between src[0..last-1], strip the funny spaces,
 344                  * updating them to tab as needed.
 345                  */
 346                 for (i = 0; i < last; i++) {
 347                         char ch = src[i];
 348                         if (ch != ' ') {
 349                                 consecutive_spaces = 0;
 350                                 strbuf_addch(dst, ch);
 351                         } else {
 352                                 consecutive_spaces++;
 353                                 if (consecutive_spaces == ws_tab_width(ws_rule)) {
 354                                         strbuf_addch(dst, '\t');
 355                                         consecutive_spaces = 0;
 356                                 }
 357                         }
 358                 }
 359                 while (0 < consecutive_spaces--)
 360                         strbuf_addch(dst, ' ');
 361                 len -= last;
 362                 src += last;
 363                 fixed = 1;
 364         } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
 365                 /* Expand tabs into spaces */
 366                 int start = dst->len;
 367                 int last = last_tab_in_indent + 1;
 368                 for (i = 0; i < last; i++) {
 369                         if (src[i] == '\t')
 370                                 do {
 371                                         strbuf_addch(dst, ' ');
 372                                 } while ((dst->len - start) % ws_tab_width(ws_rule));
 373                         else
 374                                 strbuf_addch(dst, src[i]);
 375                 }
 376                 len -= last;
 377                 src += last;
 378                 fixed = 1;
 379         }
 380
 381         strbuf_add(dst, src, len);
 382         if (add_cr_to_tail)
 383                 strbuf_addch(dst, '\r');
 384         if (add_nl_to_tail)
 385                 strbuf_addch(dst, '\n');
 386         if (fixed && error_count)
 387                 (*error_count)++;
 388 }